Following up on Karsten's SEDE query, the following is a plot of the fraction of votes that were upvotes, by month.
There isn't anything enormously surprising there. I would not expect the fraction upvotes to hold steady. It would be interesting to compare this data to the frequency of close votes.
The following shows vote totals:
Interesting is the large fluctuation from month to month, but the long-term average appears otherwise relatively steady.
I have concluded* that there are serial downvoters on the site. However, I have also concluded that serial downvoters are not necessarily pathological. The site encourages downvoting, in part because original questions become increasingly rare (at least the types of questions that are most commonly posted) and because of an excess of lower quality questions (those exhibiting minimal effort, going back to that precious term). Downvoting may be a useful reflex (see also voting as a duty below). Serial downvoting is therefore not entirely surprising.
More worrisome might be that bad actors who downvote out of spite are difficult to identify or block, and have a potentially outsized effect because of low vote counts. The problem is one of participation. Views for a question can reach 100+ within a few days, even without a question being broadcast to other sites on the network, but more usual numbers are <50. The number of people regularly voting to close and up-/downvoting are probably on the order of ~10, so cumulative votes hover around 0. Vote tallies are in general small because few people bother to vote. Question is whether most abstain because they don't care, don't have the time, or because they prefer not to cause unintended effects based on limited information.
Close votes are also decided by a dedicated cadre of fanatics. It is difficult to say whether they should be applauded or not, because to function the website needs people to perform these duties, but on the other hand leaving the fate of many questions to a small number of people cannot be good for the health of the site. Ideally a lot more people would participate in all voting activities.
Searching among heavily upvoted and unanswered questions also suggests additional explanations to voting patterns. Many heavily upvoted questions are very challenging to answer except for experts narrowly focused on very particular topics. This has multiple effects. The absence of answers for some difficult questions discourages participation. This further dilutes the number of good questions. In addition, the absence of "research-grade" questions can lead to a myopic reaction in which being answerable becomes equated with low quality or duplication.
Karsten's announcement that he is planning to deliberately increase his voting rate is perhaps a step in the right direction. In addition to reducing the statistical utility of vote tallies as measures of question quality, there might be a deflationary (cyclical feedback) effect involved in low voting rates. However, there are alternative ways of looking at this, for instance considering the "tragedy of the commons", in which some don't vote because they don't feel the need to. It makes for a fascinating topic of discussion. In the end a little common sense (such as remembering the "golden rule") and a friendly attitude is likely to benefit the site more than any profound analysis.
OTOH I think there need not be a rush to upvote or downvote in the short run if not convinced that a question is particularly good. The site does a good job of pooling opinion when questions are in fact liked, and in the long run popular posts do accumulate more votes, as it should be.
*(ok, I suspect) On various occasions (times of apparent low activity on the site) I've noticed many recent questions being downvoted - some not (imho) seeming to merit downvotes - and none receiving upvotes - even when they seemed to merit them.